DataSynthBench

A comprehensive web-based framework for synthetic data generation and automated model performance benchmarking. Upload your tabular dataset, generate synthetic variants using multiple methods, and automatically evaluate model performance drift.

🚀 Features

Core Functionality

• Multi-Method Synthetic Data Generation

  • SMOTE (Synthetic Minority Oversampling Technique)
  • GAN-based generation simulation
  • Gaussian noise injection
  • Bootstrap resampling

• Automated Model Benchmarking

  • Random Forest, Logistic Regression, SVM, XGBoost
  • Cross-validation with configurable folds
  • Performance drift detection
  • Comprehensive metric evaluation

• Interactive Dashboard

  • Real-time progress tracking
  • Visual performance comparisons
  • Drift analysis and insights
  • Export capabilities for CI/CD

Technical Features

• Modern Web Interface: React 18 + TypeScript + Tailwind CSS
• Responsive Design: Optimized for desktop and tablet workflows
• File Processing: CSV upload with automatic column type detection
• Export Formats: JSON, YAML, CSV for different use cases
• CI/CD Ready: Structured output for automated pipelines

📋 Prerequisites

Before running DataSynthBench locally, ensure you have:

• Node.js (version 16.0.0 or higher)
• npm (version 7.0.0 or higher) or yarn
• Git for cloning the repository

Check your versions:

node --version
npm --version

🛠️ Local Development Setup

1. Clone the Repository

git clone https://github.com/leandrenash/datasynthbench.git
cd datasynthbench

2. Install Dependencies

Using npm:

npm install

Using yarn:

yarn install

3. Start Development Server

npm run dev

The application will be available at http://localhost:5173

4. Build for Production

npm run build

Built files will be in the dist/ directory.

📁 Project Structure

datasynthbench/
├── public/                 # Static assets
├── src/
│   ├── components/         # React components
│   │   ├── Dashboard.tsx
│   │   ├── DatasetUpload.tsx
│   │   ├── ConfigurationPanel.tsx
│   │   ├── ResultsViewer.tsx
│   │   └── ExportPanel.tsx
│   ├── App.tsx            # Main application component
│   ├── main.tsx           # Application entry point
│   └── index.css          # Global styles
├── docs/                  # Documentation and images
├── package.json
├── README.md
└── ...config files

🎯 Quick Start Guide

1. Upload Your Dataset

• Navigate to the "Upload Data" tab
• Drag and drop a CSV file or click to browse
• Review the automatic column analysis and data preview

2. Configure Generation

• Go to the "Configure" tab
• Select synthetic data generation methods (SMOTE, GAN, Noise, Resample)
• Choose models for benchmarking
• Adjust parameters as needed

3. Run Benchmark

• Click "Run Benchmark" to start the process
• Monitor real-time progress
• View detailed results in the "Results" tab

4. Export Results

• Navigate to the "Export" tab
• Choose export format (JSON, YAML, CSV)
• Download complete results or summary reports

📊 Supported Data Formats

Input Requirements

• File Format: CSV files only
• Size Limit: Up to 100MB
• Column Types: Automatic detection of numeric and categorical columns
• Missing Values: Handled automatically during processing

Example Dataset Structure

feature1,feature2,feature3,target
1.2,category_a,0.5,class_1
2.1,category_b,0.8,class_2
1.8,category_a,0.3,class_1

⚙️ Configuration Options

Synthetic Data Generators

SMOTE

smote:
  enabled: true
  k_neighbors: 5
  sampling_strategy: "auto"

GAN Simulation

gan:
  enabled: true
  epochs: 100
  batch_size: 32

Noise Injection

noise:
  enabled: true
  noise_level: 0.1
  noise_type: "gaussian"

Resampling

resample:
  enabled: true
  strategy: "random"
  ratio: 1.0

Model Configuration

models:
  random_forest:
    enabled: true
    n_estimators: 100
  logistic_regression:
    enabled: true
    C: 1.0
  svm:
    enabled: true
    kernel: "rbf"
  xgboost:
    enabled: true
    n_estimators: 100

🔧 Development

Available Scripts

• npm run dev - Start development server
• npm run build - Build for production
• npm run preview - Preview production build
• npm run lint - Run ESLint

Code Style

• TypeScript: Strict mode enabled
• ESLint: Configured with React and TypeScript rules
• Prettier: Code formatting (recommended)

Adding New Features

1. New Synthetic Data Method:

   • Add configuration options in ConfigurationPanel.tsx
   • Implement generation logic simulation
   • Update results processing

2. New Model Type:

   • Extend model configuration interface
   • Add to benchmarking simulation
   • Update results visualization

3. New Export Format:

   • Add format option in ExportPanel.tsx
   • Implement conversion function
   • Test with sample data

🚀 Deployment

Netlify (Recommended)

1. Connect your GitHub repository to Netlify
2. Set build command: npm run build
3. Set publish directory: dist
4. Deploy automatically on push

Vercel

1. Import project from GitHub
2. Framework preset: Vite
3. Build command: npm run build
4. Output directory: dist

Docker

FROM node:18-alpine
WORKDIR /app
COPY package*.json ./
RUN npm ci --only=production
COPY . .
RUN npm run build
EXPOSE 3000
CMD ["npm", "run", "preview"]

🤝 Contributing

We welcome contributions! Please see our Contributing Guidelines for details.

Development Workflow

1. Fork the repository
2. Create a feature branch: git checkout -b feature/amazing-feature
3. Make your changes
4. Run tests: npm run lint
5. Commit changes: git commit -m 'Add amazing feature'
6. Push to branch: git push origin feature/amazing-feature
7. Open a Pull Request

Reporting Issues

• Use the GitHub Issues page
• Include detailed reproduction steps
• Provide sample data if possible (anonymized)

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🗺️ Roadmap

• Python backend integration for real synthetic data generation
• Advanced visualization with D3.js
• Model explainability features
• Automated hyperparameter tuning
• Integration with MLflow and other ML platforms
• Support for time series data
• Advanced drift detection algorithms

---

Made with ❤️ for the data science community